Скачать или смотреть Jose Walteros - A branch-and-bound framework to solve interdependent multi-stage interdiction games

Part of Discrete Optimization Talks: https://talks.discreteopt.com

Jose Walteros - University at Buffalo

Speaker webpage: https://researchgamma.com/Jose_Direct...

A branch-and-bound framework to solve a general class of interdependent multi-stage interdiction games

Abstract: We study a general class of three-stage interdiction games that model the sequential interaction of three decision-makers: a protector, an attacker, and a defender. In the first stage, the protector conducts a set of fortification strategies to protect from the attacker the defender's course of actions; in the second stage, the attacker deploys a set of disruptive attacks, parameterized by the fortification strategies, to optimally deteriorate the operators' objective; finally, in the third stage, the operator optimizes their course of action, considering the effects of the previous stages. We focus on a general variation in which the actions taken by the protector can directly affect the solution space of the defender. In other words, the protection actions made in the first stage may also turn solutions for the defender infeasible. General applications of this type of game arise in problems with an underlying network structure, where the protector can alter the composition of the network by adding/removing edges and vertices. In this presentation, we develop an exact solution framework for these problems based on a combinatorial branch-and-bound search. We provide some examples of applications arising from communications and network design problems.

Bio: Dr. Jose L. Walteros is an Assistant Professor in Industrial and Systems Engineering at the University at Buffalo. He holds a Ph.D. in Industrial and Systems Engineering from the University of Florida. Dr. Walteros has extensive experience in developing exact and approximate solutions for solving challenging problems in a wide variety of areas, including attacker-defender games, logistics, transportation and routing, shared-mobility systems management, homeland security, survivable network design, data association, and social networks analysis, with 25+ research publications in these fields. His research has been funded by organizations such as the National Science Foundation (NSF), the Office of Naval Research (ONR), the Air Force Research Laboratory (AFOSR), and the Department of Transportation (DOT), among others. Previously, he served as the President of the INFORMS Junior Faculty Interest Group and the Vice Chair for Network Optimization in the INFORMS Optimization Society (IOS). His papers have appeared in the journals Operations Research, INFORMS Journal on Computing, Networks, Mathematical Programming, Mathematical Programming Computation, Transportation Science, the European Journal of Operational Research, and others.